Risk of second cancer after chronic lymphocytic leukemia

Authors


Abstract

Smoking is not considered a risk factor for chronic lymphocytic leukemia (CLL) yet increased lung cancer risk has been reported for these patients. Little data exist on the temporal variation in lung cancer risk after CLL, or its histological composition. We investigated the occurrence of second cancers in a large cohort of CLL patients with particular emphasis on lung cancer and its major subtypes. We followed all patients diagnosed with CLL in Denmark in the period 1943–2003 (n = 12,373) for the occurrence of second cancers. The relative risk was expressed as the standardized incidence ratio (SIR), i.e. the ratio of observed to expected number of cancers, based on incidence rates for the Danish population. During follow-up 1,105 cancers occurred among the CLL patients (SIR = 1.59 (95% CI 1.50–1.69)). SIR for all cancers combined remained elevated more than 10 years after CLL (SIR = 1.80 (1.56–2.08)). Lung cancer occurred in 141 patients (SIR = 1.61 (1.37–1.90)). The relative risk of lung cancer did not vary by gender, or time of follow-up, but was higher in younger (SIR<60 years = 2.22 (1.62–3.06)) than in older (SIR70–79 years = 1.21 (0.88–1.68)) age-groups. Elevated risks were observed for adenocarcinoma (SIR = 2.20 (1.57–3.08)) and squamous cell carcinoma (SIR = 1.52 (1.06–2.17)) of the lung. We speculate that shared genetic risk factors may explain the accumulation of lung and other cancers in CLL patients. © 2007 Wiley-Liss, Inc.

In Western countries chronic lymphocytic leukemia (CLL) accounts for 25–30% of all leukemias.1 Clinically the disease varies substantially: while in many patients it remains indolent with no need for treatment, in others it displays an aggressive clinical course. Consequently the 7-years median survival time for CLL ranges from a few months to several decades.2

The causes of CLL largely have remained unestablished,3, 4, 5 but there is increasing evidence that its development involves both constitutional and environmental factors.6, 7 Little is known about the nature of these factors, but important clues could emanate from studies of comorbidity in CLL patients, e.g. the occurrence of second cancers.

Previous studies have indicated that patients with CLL have an increased risk of subsequent cancer, in particular cancers of the skin and lung.8, 9, 10, 11, 12 Several mechanisms have been discussed for the association between these cancers. The increased risk of lung cancer is particularly noteworthy, because smoking has so far not been linked with CLL risk.13, 14 In previous Danish11 and American10 studies, the increased risk of lung cancer was found to be independent of time since CLL diagnosis, leading to the conclusion that it could not be attributed to CLL treatment offered no explanation for the observed increased lung cancer risk. However, concomitant with the introduction of new treatment strategies including the nucleoside analog Fludarabine, other studies have reported both increased risk of lung cancer,15, 16, 17 and acute myeloid leukemia,16, 18 and a more aggressive course of second malignancies.19, 20

We took advantage of long-standing Danish registers to update and further characterize the risk of second cancers in patients with CLL. In particular, as a novelty we evaluated the occurrence of secondary lung cancers by their histological subtypes.

Material and methods

Patients

The Danish Cancer Register contains information on all incident cancers in Denmark since 1943.21 Details of individual cases of cancer are available according to the 7th revision of the International Classification of Diseases (ICD) for all years, and according to the ICD-O since 1978. In the Register, we identified all patients with chronic lymphocytic leukemia (ICD-7-code 204.0) and all other cancers registered for these individuals. The information available for each individual among others included date of birth, sex, and dates of cancer diagnoses. Information on the patients' vital status was obtained from the Danish Civil Registration System, except for patients who had died before April 1, 1968 for whom date of death was obtained in the Cancer Register. For the present analyses, cases of lung cancer among the CLL patients diagnosed after 1978 were categorized into the following subtypes (ICD-7-code 162.0, 162.1 and 162.4–162.7): adenocarcinoma, large-cell carcinoma, small-cell carcinoma, squamous cell carcinoma and unclassified carcinoma (all other histological groups). The study was approved by the Danish Data Protection Agency (Reference no. 2002-41-2066).

Statistical analysis

Follow-up of the CLL patients began on the month following CLL diagnosis for analyses of overall cancer by time since CLL, and 1 year after CLL diagnosis in all other analyses to reduce the impact of potential diagnostic misclassification. Follow-up ended on the date of death, emigration or December 31, 2003. The relative risk of cancer was expressed as the ratio of observed to expected number of cancers, i.e. the standardized incidence ratio (SIR). The expected number of cancers was calculated by multiplying age-, sex- and calendar period-specific cancer incidence rates in the background population by correspondingly stratified person-years at risk in the CLL cohort, and summing all the products. Analyses were carried out for all CLL patients combined and stratified according to gender and age, specifically. Individual treatment information was not available. However, to investigate the potential impact of treatment with the nucleoside analogue Fludarabin on second cancer risk, we divided the follow-up period into eras before and after September 1994, when the compound became available for routine treatment in Denmark. The 95% confidence intervals (CI) for the SIRs based on Wald's test assuming a Poisson distribution of the observed events.

Results

Overall, 12,373 patients diagnosed with CLL between 1943 and 2003 were identified (Table I). The median age at CLL onset was 70 years and the male-to-female ratio was 1.6:1. The CLL patients were followed for 47,636 person-years at risk (mean time of follow-up 3.8 years) during which 1,105 cancers occurred. SIR for all cancers combined was 1.59 (95% CI 1.50–1.69), ranging from 1.63 (95% CI 1.44–1.85) in the first year after CLL to 1.80 (95% CI 1.56–2.08) in the period more than 10 years after CLL diagnosis. The relative risk for cancer overall was slightly higher in men (SIR = 1.74 (95% CI 1.62–1.87)) than in women (SIR = 1.36 (95% CI 1.23–1.50); phomogeneity = <0.0001). This gender difference was also seen across strata of age and length of follow-up (Table II). Younger patients with CLL seemed to be at a higher relative risk of second cancer than older patients (Table II).

Table I. Demographic Characteristics of Patients Diagnosed with Chronic Lymphocytic Leukemia From 1943 Till 20031
 Patients (%)Person-years (%)
  • 1

    Based on data from the Danish Cancer Register.

Overall12,373 (100)47,636 (100)
Gender
 Male7,551 (61)26,632 (56)
 Female4,822 (39)21,004 (44)
Age at diagnosis of CLL (in years)
 0–592,484 (20)13,692 (29)
 60–693,346 (27)15,224 (32)
 70–794,196 (34)14,078 (30)
 ≥802,347 (19)4,642 (10)
Calendar period of diagnosis of CLL
 1943–19591,724 (14)4,532 (9)
 1960–19691,450 (12)5,072 (11)
 1970–19792,050 (17)8,048 (17)
 1980–19892,771 (22)13,086 (27)
 1990–19993,108 (25)14,710 (31)
 2000–20031,270 (10)2,186 (5)
Table II. Relative Risk of Second Cancer in Patients Diagnosed with Chronic Lymphocytic Leukemia from 1943 Till 2003
 All patientsMenWomen
ObservedExpectedSIR (95% CI)ObservedSIR (95% CI)ObservedSIR (95% CI)
  • 1

    Excluding cancers diagnosed less than one year after CLL.

  • 2

    Test for homogeneity of relative risk of second cancer between genders.

  • 3

    Test for homogeneity between genders adjusted for age at diagnosis of CLL. Test for interaction between gender and age at diagnosis of CLL pinter = 0.0098.

  • 4

    Test for homogeneity between time-periods since CLL diagnosis, exclusive the first year after CLL diagnosis.

  • 5

    Test for homogeneity between genders adjusted for time since CLL.

All second cancers1
 1943–20031,105695.11.59 (1.50–1.69)7341.74 (1.62–1.87)3711.36 (1.23–1.50)
       Phom = <0.00012
 1943–September 1994630388.01.62 (1.50–1.76) 
 October 1994–2003475307.11.55 (1.41–1.69) 
   Phom = 0.27 
Age in years at diagnosis of CLL1
 0–59233112.52.07 (1.82–2.36)1752.49 (2.15–2.89)581.38 (1.06–1.78)
 60–69385228.01.69 (1.53–1.87)2571.83 (1.62–2.06)1281.47 (1.23–1.75)
 70–79392269.61.45 (1.32–1.61)2561.55 (1.37–1.75)1361.30 (1.10–1.54)
 ≥809585.11.12 (0.91–1.36)461.00 (0.75–1.33)491.26 (0.95–1.66)
   Phom = <0.0001   Phom = <0.00013
Time since CLL diagnosis
 0 year241147.51.63 (1.44–1.85)1701.75 (1.51–2.04)711.40 (1.11–1.77)
 1–4 years578391.31.48 (1.36–1.60)3851.55 (1.40–1.71)1931.35 (1.17–1.56)
 5–9 years338198.81.70 (1.53–1.89)2311.96 (1.72–2.23)1071.32 (1.09–1.60)
 ≥10 years189105.01.80 (1.56–2.08)1182.11 (1.76–2.53)711.45 (1.15–1.83)
   Phom = 0.0244   Phom = <0.00015

Statistically significantly increased relative risks were observed for cancers at several anatomical sites, including the respiratory system (Table III). Decreased risk was observed for cancers of the breast and the female genital organs (Table III).

Table III. Relative Risk of Second Cancer in Patients Diagnosed with Chronic Lymphocytic Leukemia From 1943 Till 2003
ICD-7Cancer groupNumber observedNumber expectedSIR (95% CI)
140–250All malignant neoplasms1,105695.11.59 (1.50–1.69)
140–148Buccal cavity and pharynx2313.01.77 (1.18–2.67)
 140Lip64.01.50 (0.67–3.34)
 141Tongue21.61.27 (0.32–5.10)
 142Salivary glands41.23.36 (1.26–8.95)
 143–144Mouth63.31.81 (0.81–4.03)
 145–148Pharynx52.91.72 (0.72–4.14)
150–159Digestive organs and peritoneum188181.21.04 (0.90–1.20)
 150Esophagus118.21.34 (0.74–2.41)
 151Stomach2527.80.90 (0.61–1.33)
 152Small intestine01.7
 153Colon, inclusive recto sigmoid6664.61.02 (0.80–1.30)
 154Rectum, exclusive anus3435.10.97 (0.69–1.35)
 155.0Liver66.50.92 (0.41–2.04)
 155.1Gallbladder till amp. Vateri76.71.05 (0.50–2.20)
 156Liver, not specified as primary88.20.97 (0.49–1.94)
 157Pancreas2720.71.31 (0.90–1.91)
 158–159Peritoneum and unspecified41.62.45 (0.92–6.51)
160–164Respiratory system16297.91.66 (1.42–1.93)
 160Nasal cavities, and sinuses21.41.39 (0.35–5.56)
 161Larynx95.91.54 (0.80–2.96)
 162.0–162.1Lung primary, tracheae14187.61.61 (1.37–1.90)
 162.2Pleura72.13.27 (1.56–6.85)
 163Lung, not specified as primary10.52.06 (0.29–14.6)
 164Mediastinum20.45.33 (1.33–21.3)
170Breast3550.00.70 (0.50–0.97)
171–176Female genital organs2031.10.64 (0.41–1.00)
 171Cervix uteri46.00.67 (0.25–1.78)
 172Corpus uteri711.30.62 (0.30–1.30)
 173–174Uterus, other parts and unspecified00.7
 175Ovary, fallopian tube, broad ligament810.20.78 (0.39–1.57)
 176Other and unspecified female genital organs12.90.35 (0.05–2.45)
177–179Male genital organs6164.30.95 (0.74–1.22)
 177Prostate5762.00.92 (0.71–1.19)
 178Testis20.82.55 (0.64–10.2)
 179Other and unspecified male genital organs21.51.33 (0.33–5.31)
180–181Urinary system7864.61.21 (0.97–1.51)
 180Kidney3318.01.84 (1.31–2.58)
 181Bladder incl. papilloma4546.70.96 (0.72–1.29)
190–191Skin430121.23.55 (3.23–3.90)
 190Melanoma of skin2711.22.42 (1.66–3.53)
 191Other skin403110.13.66 (3.32–4.04)
192–197Other specified sites2418.41.30 (0.87–1.94)
 192Eye21.31.52 (0.38–6.06)
 193Brain and nervous system1112.30.89 (0.49–1.61)
 194Thyroid51.92.64 (1.10–6.35)
 195Endocrinal glands00.5
 196Bone20.54.26 (1.07–17.0)
 197Connective tissue42.02.04 (0.77–5.43)
198–199Secondary and unspecified sites2921.41.35 (0.94–1.95)
 198Metastases1310.21.27 (0.74–2.19)
 199Other and unspecified sites1611.21.43 (0.88–2.33)
200–205Lymphatic and haematopoetic tissue5532.01.72 (1.32–2.24)
 200, 202Non-Hodgkin lymphoma3814.02.73 (1.99–3.75)
 201Hodgkin lymphoma71.44.95 (2.36–10.4)
 203Multiple myeloma48.20.49 (0.18–1.31)
 204Leukemia58.20.61 (0.26–1.47)
 205Mycosis fungoides10.42.66 (0.37–18.9)

Lung cancer occurred in 141 patients (110 men, 31 women), corresponding to an SIR of 1.61 (95% CI 1.37–1.90) (Table IV). The relative risk of lung cancer did not differ between women and men, but varied slightly by age at CLL (Table IV). The increased relative risk of lung cancer could be assumed to be uniformly distributed over different periods of follow-up (Table IV).

Table IV. Relative Risk of Second Lung Cancer in Patients Diagnosed with Chronic Lymphocytic Leukemia From 1943 Till 2003
 All patientsMenWomen
ObservationsSIR (95% CI)ObservationsSIR (95% CI)ObservationsSIR (95% CI)
  • 1

    Excluding lung cancers, diagnosed less than one year after CLL.

  • 2

    Test for homogeneity of relative risk of lung cancer between genders.

  • 3

    Test for homogeneity between genders adjusted for age at diagnosis of CLL. Test for interaction between gender and age at diagnosis of CLL pinter = 0.83.

  • 4

    Test for homogeneity between time-periods since CLL diagnosis, exclusive the first year after CLL diagnosis.

  • 5

    Test for homogeneity between genders adjusted for time since CLL, exclusive the first year after CLL diagnosis.

All second lung cancers1
 1943–2003141161 (1.37–1.90)1101.65 (1.37–1.99)311.49 (1.05–2.12)
      Phom = 0.622
Age in years at diagnosis of CLL1
 0–59382.22 (1.62–3.06)312.35 (1.66–3.35)71.79 (0.85–3.75)
 60–69591.76 (1.36–2.27)471.83 (1.38–2.44)121.51 (0.86–2.66)
 70–79371.21 (0.88–1.68)271.16 (0.79–1.69)101.41 (0.76–2.62)
 ≥8071.09 (0.52–2.28)51.09 (0.45–2.62)21.08 (0.27–4.34)
  Phom = 0.037   Phom = 0.653
Time since CLL diagnosis
 0 year381.98 (1.44–2.73)332.13 (1.52–3.00)51.36 (0.57–3.27)
 1–4 years791.56 (1.25–1.94)611.53 (1.19–1.96)181.65 (1.04–2.62)
 5–9 years471.90 (1.43–2.53)402.17 (1.59–2.95)71.11 (0.53–2.34)
 ≥10 years151.25 (0.75–2.07)91.07 (0.56–2.06)61.66 (0.75–3.70)
  Phom = 0.314   Phom = 0.635

Information on histological lung cancer subtypes was available for cancer diagnosed after 1977 and included 119 cases (88 men, 31 women). Detailed analyses showed statistically significantly elevated relative risks for adenocarcinoma and squamous cell carcinoma, and nonsignificantly elevated estimates for large-cell and small-cell carcinomas. The relative risk of lung cancer subtypes was evenly increased in men and women, and across all strata of age at CLL and time since CLL (Table V).

Table V. Patients with Chronic Lymphocytic Leukemia, Diagnosed with Second Lung Cancer1 from 1978 Till 2003, Stratified by Lung Cancer Subtypes
 AdenocarcinomaLarge-cell carcinomaSmall-cell carcinomaSquamous cell carcinomaUnclassified
No.SIR (95% CI)No.SIR (95% CI)No.SIR (95% CI)No.SIR (95% CI)No.SIR (95% CI)
  • No. = Number of observed second cancers.

  • 1

    Registration of histological lung cancer subtypes since 1978.

  • 2

    Excluding lung cancers, diagnosed less than one year after CLL.

  • 3

    Test for homogeneity of relative risk between lung cancer subtypes exclusive unclassified lung cancer (inclusive unclassified lung cancer phom = 0.086).

  • 4

    Test for homogeneity of relative risk of lung cancer subtypes between genders.

  • 5

    Test for homogeneity between time-periods since CLL diagnosis, exclusive the first year after CLL diagnosis.

All patients2342.20 (1.57–3.08)51.44 (0.60–3.45)141.06 (0.63–1.78)301.52 (1.06–2.17)361.73 (1.24–2.39)
  Phom = 0.0563
 Men252.45 (1.65–3.62)41.56 (0.59–4.16)101.03 (0.55–1.91)211.25 (0.81–1.92)281.83 (1.27–2.66)
 Women91.72 (0.89–3.30)11.09 (0.15–7.76)41.13 (0.42–3.01)93.02 (1.57–5.81)81.43 (0.72–2.86)
  Phom = 0.764 Phom = 0.464 Phom = 0.614 Phom = 0.244 Phom = 0.404
Age in years at diagnosis of CLL2
 0–59123.02 (1.72–5.32)11.25 (0.18–8.87)82.67 (1.33–5.33)61.49 (0.67–3.32)72.74 (1.31–5.75)
 60–69202.70 (1.74–4.19)31.80 (0.58–5.57)60.93 (0.42–2.06)181.88 (1.18–2.98)162.18 (1.34–3.56)
 70–79122.00 (1.14–3.52)21.41 (0.35–5.64)50.87 (0.36–2.10)151.69 (1.02–2.80)121.04 (0.59–1.84)
 ≥8010.89 (0.13–6.35)13.46 (0.49–24.59)0032.07 (0.67–6.41)41.15 (0.43–3.06)
  Phom = 0.45 Phom = 0.90 Phom = 0.077 Phom = 0.95 Phom = 0.11
Time to second malignancy
 0 year113.64 (2.01–6.57)22.86 (0.72–11.45)51.83 (0.76–4.39)122.88 (1.64–5.07)30.75 (0.24–2.33)
 1–4 years171.95 (1.21–3.13)52.53 (1.05–6.08)91.18 (0.61–2.27)181.57 (0.99–2.49)161.40 (0.86–2.29)
 5–9 years112.43 (1.35–4.39)030.79 (0.25–2.44)81.41 (0.71–2.83)172.75 (1.71–4.42)
 ≥10 years62.72 (1.22–6.05)021.10 (0.27–4.39)41.50 (0.56–4.00)30.92 (0.30–2.84)
  Phom = 0.475   Phom = 0.705 Phom = 0.355 Phom = 0.0505

We also estimated the relative risk of cancer before and after 1994 to indirectly evaluate the possible effect of Fludarabine treatment. While the relative risk of all second cancers combined did not differ (SIR = 1.62 (95% CI 1.50–1.76) before 1994, and SIR = 1.55 (1.41–1.69) after 1994; phomogeneity = 0.27), lower relative risks were observed after 1994 for lung cancer (SIR = 1.29 (0.97–1.71)), prostate cancer (SIR = 0.63 (0.40–1.00)) and for all hematopoietic and lymphatic cancers combined (SIR = 1.40 (1.03–1.91)), than before this date (SIRs = 1.85 (1.51–2.26), 1.16 (0.85–1.59), 1.97 (1.56–2.49), respectively; phomogeneity = 0.04, 0.03, 0.04, respectively). In contrast, a higher relative risk was observed for breast cancer after (SIR = 0.91 (0.68–1.24)) than before 1994 (SIR = 0.52 (0.36–0.75); phomogeneity = 0.02). For all other anatomical sites cancer risk remained unchanged (data not shown).

Because the Cancer Registry may not have been complete in the years 2002–2003 for patients whose cancer diagnoses rested exclusively on death certificates information (constituting less than 1% of all registrations in the Cancer Register in 2001), all analyses were repeated for the period until 2001, reassuringly yielding results essentially identical to those presented.

Discussion

We observed an increased risk of second cancers in patients with CLL, consistent with previous observations (reviewed in Refs.22,23). The increased cancer risk remained relatively constant over time since diagnosis of CLL and was observed in both men and women, although men were at significantly higher relative risk than women.

Increased risk of skin cancer, melanoma as well as nonmelanoma, has previously been described in CLL patients.24 With 430 observed cases corresponding to a more than 3-fold increased risk, it was the most frequent second cancer site in our study. UV light has been suggested as a possible shared risk factor for CLL and skin cancer.25 However, in a recent study of more than 3,700 cases of Swedish and Danish patients with non-Hodgkin lymphoma (NHL) no association between UV light exposure and CLL or other NHL subtypes was observed.26 Given the typically benign course of nonmelanoma skin cancer and the suspected incomplete registration of this cancer form in general,27 we cannot exclude the possibility that the observed increased risk to some extent reflects surveillance and registration bias. This explanation, however, hardly applies to the increased occurrence of malignant melanomas, which therefore remains unexplained.

In the present analyses, elevated risks were observed for several cancer sites for which smoking has a well-established etiological role.28 This included cancers of the buccal cavity and pharynx, esophagus, pancreas, the respiratory system and the urinary systems. This is remarkable because there is little evidence for an association between smoking and CLL.14 Smoking data were not available in our register-based study, however, in a previous study of CLL and second malignancies similar prevalence of smoking was observed in the CLL patients (84%) and the general population (80%), leading the authors to conclude that the increased second cancer risk among the CLL patients group was hardly explained by their smoking habits.15

Cigarette smoking is the cause of 85–90% of cases of lung cancer, and while it cannot be ruled out that its carcinogenicity is augmented in CLL due to radiotherapy or hypothesized immune perturbations,10 the cancer has also been linked with certain environmental and occupational exposures, i.e. asbestos and radon.29, 30 Additionally, it has been suggested that the etiology of lung cancer is multifactorial, and involves both environmental and genetic factors,31 possibly acting in synergy. Several genetic factors influencing the susceptibility of developing both CLL and lung cancer have been discussed. Of particular interest, genetic polymorphisms in enzymes involved in the metabolism of tobacco-specific carcinogens may modify the risk for both CLL and lung cancer. Specifically, polymorphisms in glutathione S-transferase genes, encoding for a polycyclic aromatic hydrocarbon involved in the metabolism of tobacco carcinogens, seem to modulate the risk of smoke-related cancer, especially lung cancer,32, 33, 34 but also CLL.35

Risk factors for lung cancer presumably apply to different degrees to the major subtypes. Thus, of the different subtypes, squamous cell carcinoma seems to be the strongest associated with smoking,36 whereas large-cell and adenocarcinoma cluster the most in families.37 Although we note that the observed relative risk estimate was the highest for adenocarcinomas in our analyses, the rather even distribution of relative risks for various lung cancer subtypes in the CLL patients offered no further insight into putative mechanisms for the association.

We observed an increased relative risk of NHL (SIR = 2.73), and of Hodgkin lymphoma (SIR = 4.95) after CLL. This could be due the evolvement of CLL into aggressive large-cell lymphoma (Richter's transformation/syndrome), often resembling Hodgkin lymphoma.1, 17 The risk of thyroid cancer was increased (SIR = 3.04). Although the mechanism is not clearly established, it is interesting that CLL has been associated with autoimmune thyroiditis,38, 39 which in turn has been linked with thyroid cancer.40 A reduction in risk of breast cancer and endometrial malignancies has been noted in several studies of NHL patients.41, 42 This has been explained by the induction of premature menopause caused by ovarian failure following treatment with cyclophosphamide. We observed a decreased risk of cancers of the breast (SIR = 0.70), and of the female organs (SIR = 0.64) among CLL patients. Although cyclophosphamide has been standard chemotherapy for CLL for several decades,43 the same mechanism can hardly explain our findings since the majority of female CLL patients were postmenopausal. While obesity has been associated with an increased risk of both endometrial cancer,44 and certain subtypes of postmenopausal breast cancer,45 no compelling association between height and weight and CLL has been reported so far.46, 47

The influence of chemotherapy treatment, overall or in regiments using combinations of different agents, on the risk of second cancers in CLL is controversial. Specifically, some studies with individual treatment information have indicated a more aggressive second cancer phenotype after treatment with the nucleoside analog Fludarabine,48, 49 others that the treatment with the nucleoside analog Cladribine may further augment the increased second cancer risk in CLL patients,17 and still others that Fludarabine is of no consequence to the increased second cancer occurrence.50 In Denmark, Fludarabine was registered for CLL treatment in October 1994. To evaluate possible contemporary changes in cancer risk, we compared second cancer risks in CLL patients before and after this date. While the estimate of relative cancer risk overall did not differ, estimates for the relative risk of lung cancer, if anything, was lower among CLL patients after September 1994 than before this date, as were the relative risks of prostate cancer and of all hematopoietic and lymphatic cancers combined. The unchanged cancer risk with potential Fludarabine exposure is similar to observations by Hisada et al.10 who also observed no variation in second cancer risk between with (1990–1996) or without (1973–1989) potential nucleoside analog exposure. While these results would suggest that Fludarabine does not explain or affect the increased risk of lung and other cancers after CLL, it should be recognized that the influence of other factors, e.g. changes in smoking habits cannot be ruled out in our analyses. Accordingly, further studies of the impact of nucleoside analogues seem warranted.

Hisada et al. assessed second cancer risk in 16,367 CLL patients registered in the surveillance, epidemiology and end results (SEER) program between 1973 and 1996.10 Compared with the present investigation, they observed a slightly less increased relative risk of cancer overall (SIR = 1.20 (95% CI 1.15–1.26); n = 1,820), which consistent with our findings persisted more than 10 years after CLL diagnosis. With respect to anatomical sites, increased risks were observed for lung cancer (SIR = 1.66 (p < 0.05), similar in men and women) and for malignant melanoma (SIR = 3.18 (p < 0.05)) in the SEER study like in most other investigations. Interestingly, Hisada et al. also observed an excess of Kaposi sarcomas among the CLL patients (SIR = 5.09 (p < 0.05)) based on 9 observed cases, fostering speculations about the role of CLL-associated immunological impairment. No cases of Kaposi sarcoma were observed among CLL patients in our investigation, and similarly, no cases of CLL were observed in a follow-up study of 741 Scandinavian patiens with classical Kaposi sarcoma.51 We cannot readily explain this discrepancy between our findings and those of Hisada et al.,10 but speculate that it may be attributable to chance or diagnostic misclassification or registration phenomena.

Limitations of our study include missing information on smoking habits and chemotherapy treatment. Also the classification of malignant lymphomas has changed during the study period, most recently leading to small lymphocytic lymphoma being recognized as CLL in the WHO classification.52 In contrast, the strengths are cohort data based on the Danish Cancer Register records of all cases of cancer in Denmark since 1943, and available information on vital status of these patients by linkage of the Danish Cancer Register with the Danish Civil Registration System.

To summarize, we observed an increased occurrence of lung cancer, in particular adenocarcinomas and squamous cell carcinomas, and of other smoking related malignancies among CLL patients. The increased risk of cancer was uniformly distributed across all periods of follow-up, suggesting that it was unrelated to initial CLL therapy and not explained by diagnostic misclassification (initially) or surveillance bias (later). Tobacco smoking is the major risk factor in lung cancers, however, little evidence of an association between smoking and CLL is available from previous studies. Family aggregation found for both lung cancer and CLL may indicate hereditary predisposition to the two conditions, and accordingly, shared genetic risk factors could explain the accumulation of lung cancer, but possibly also other second cancers, in CLL patients. Our findings add further to the growing evidence of an increased risk of subsequent cancers in CLL patients. This should be relevant to physicians treating CLL patients with respect to information of the patients concerning lifestyle, e.g. tobacco smoking and sun exposure, and to continued cancer surveillance, e.g. by adherence to cancer screening programs.

Ancillary